# NAME

Interpolation - Arbitrary string interpolation semantics (using tie())

Version 0.70

Originaly by Mark-Jason Dominus (mjd-perl-interpolation@plover.com) Since version 0.66 maintained by Jenda@Krynicky.cz

# SYNOPSIS

use Interpolation name => \&function, ...;
print "la la la la \$name{blah blah blah}";

# This is like doing:
\$VAR = &function(blah blah blah);
print "la la la la \$VAR";

# DESCRIPTION

Beginners always want to write this:

print "The sum of three and four is: 3+4";

And they want the 3+4 part to be evaluated, so that it prints this:

The sum of three and four is: 7

Of course, it's a double-quoted string, so it's not evaluated. The only things that are evaluated in double-quoted strings are variable references.

There are solutions to this, but most of them are ugly. This module is less ugly. Well .... this module IS ugly, but only inside. Your code may end up being nice.

The module also lets you define arbitrary interpolation semantics.

For example, you can say

use Interpolation money => \&commify_with_dollar_sign,
E     => 'eval',
placename => 'ucwords',
;

And then you can write these:

print "3 + 4 = \$E{3+4}";
# Prints  ``3 + 4 = 7''

\$SALARY = 57500;
print "The salary is \$money{\$SALARY}";
# Prints  ``The salary is \$57,500.00''

\$PLACE1 = 'SAN BERNADINO HIGH SCHOOL';
\$PLACE2 = 'n.y. state';
print "\$placename{\$PLACE1} is not near \$placename{\$PLACE2}";
# Prints  ``San Bernadino High School is not near N.Y. State";

# DETAILS

The arguments to the use call should be name-function pairs. If the pair is (\$n, \$f), then \$n will be the name for the semantics provided by \$f. \$f must either be a reference to a function that you supply, or it can be the name of one of the built-in formatting functions provided by this package. Interpolation will take over the %n hash or \$n scalar in your package, and tie it so that acessing \$n{X} calls f(X) and yields its return value.

If for some reason you want to, you can add new semantics at run time by using

import Interpolation name => function, ...

You can remove them again with

unimport Interpolation 'name', ...

Interpolators are always PACKAGE variables, not lexicals!

## Built-ins

Interpolation provides a few useful built-in formatting functions; you can refer to these by name in the use or import line. They are:

eval                    - Evaluate the argument
null                    - Same as eval
identity                - Also the same as eval

ucwords         - Capitalize Input String Like This

commify         - 1428571 => 1,428,571.00

reverse         - reverse string

sprintf                 - makes "\$S{'%.2f %03d'}{37.5,42}" turn into "37.50 042"
use Interpolation S => 'sprintf';
print "\$S{'%.2f %03d'}{37.5, 42}\n";

sprintf1                - makes "\$S{'%.2f %03d', 37.5,42}" turn into "37.50 042".
use Interpolation S => 'sprintf1';
print "\$S{'%.2f %03d', 37.5, 42}\n";

sprintfX                - makes "\$S{'%.2f %03d'}{37.5}{42}" turn into "37.50 042".
use Interpolation 'S:\$\$*->\$' => 'sprintfX';
print "\$S{'%.2f %03d'}{37.5}{42}\n";

sqlescape               - escapes single quotes for use in SQL queries

round                   - rounds the number
use Interpolation round => 'round'; print "The sum is: \$round{\$sum, 0.01}\n";
use Interpolation 'round:\$\$->\$' => 'round'; print "The sum is: \$round{\$sum}{0.01}\n";

htmlescape      - escapes characters special to HTML
"<b>\$htmlescape{\$text}</b>
tagescape       - escapes characters special to HTML plus double and single quotes
qq{<input type=text name=foo value="\$tagescape{\$value}">}
jsescape                - escapes the text to be used in JavaScript
qq{<a href="JavaScript:foo( '\$jsescape{\$value}' )">}
(the last three require module HTML::Entities)

It is posible to pass multiple (or no) arguments to your function. There are two alternate syntaxes:

\$interpolator{param1,param2}
\$interpolator{param1}{param2}

The first syntax will pass both arguments in \$_[0] joined by \$;, so you have to split them:

use Interpolation add => sub{@_ = split /\$;/o, \$_[0]; \$_[0] + \$_[1]};
print "3 + 4 = \$add{3,4}\n";

The other syntax (used for example by builtin 'sprintf') requires quite some magic, so you probably wouldn't want to be forced to write it yourself. (See the source of this module if you want to know how strange is the code. )

The other problem is, that your interpolator might want to return an array. In that case you would anticipate to get all the items joined by \$", but instead you would get only the last item. You have to join the list yourself:

use Interpolation foo => sub {join \$", &bar(\$_[0])};

To make your life easier this module provides a way to specify the "type" of the interpolator and then does the necessary splits, joins or magic itself.

The syntax is:

use Interpolation 'name:input->output' => sub { ...

where the input is a list of '\$'s, '@'s and '\@'s and the output is either '\$' or '@'. The '\$' means that the parameter/output should be left intact, while '@' forces a split/join on the parameter/output. Each character in the input list specifies the type of one brace in the call.

In addition you may add an asterisk to the end of the input type specification. This will allow for an arbitrary long list of parameters. Their type will be the last specified.

In previous version you had to "close" the interpolator call by \$;. That is you would write something like "xxx \$foo{par1}{par2}...{parn}{\$;} xxx". While this is still suported it is NOT required anymore.

The default type is \$->\$.

Ex.:
'foo:\$->\$' - pass the argument to function directly and evaluate it in scalar context
\$foo{param}
'foo:\$->@' - pass the argument to function directly, evaluate it in list context and join
the result by \$" (by default space)
\$foo{param}
'foo:@->\$' - split the first parameter by \$; and pass the resulting list to the function,
evaluate in scalar context
\$foo{param1,param2,...}
'foo:@->@' - split the first parameter by \$; and pass the resulting list to the function,
evaluate in list context and join
\$foo{param1,param2,...}
'foo:\$\$->\$' - ask for two parameters enclosed in braces
\$foo{param1}{param2}
'foo:\$@->\$' - ask for two parameters enclosed in braces and split the second one
the list you get from the split will be added to @_ flatlist
\$foo{paramA}{paramB1,paramB2,...}
'foo:\$\@->\$' - ask for two parameters enclosed in braces and split the second one
the list you get from the split will be passed as a reference to an array
\$foo{paramA}{paramB1,paramB2,...}
'foo:\$*->\$   - ask for arbitrary number of scalar parameters
\$foo{par1}{par2}{par3}{\$;}
'foo:->\$      - no parameters. This creates a tied scalar.
\$foo

'foo:@->\$' => &bar   IS EQUAL TO   'foo' => sub {&bar(split /\$;/o, \$_[0])}
'foo:\$->@' => &bar   IS EQUAL TO   'foo' => sub {join \$", &bar(\$_[0])}
'foo:@->@' => &bar   IS EQUAL TO   'foo' => sub {join \$", &bar(split /\$;/o, \$_[0])}
'foo:\@->\$' => &bar  IS EQUAL TO   'foo' => sub {&bar([split /\$;/o, \$_[0] ])}

The builtin function sprintf could be implemented as: 'sprintf:\$@->\$' => sub {sprintf shift,@_}

Since version 0.69 it is possible to assign to interpolators of type '\$->\$', '\$->@' and '->\$'. The assigned value will be passed to the function you specified as the last parameter:

use Interpolation 'count:->\$' => sub {if (@_) {\$count = \$_[0]} else {\$count++}};
# print "Current count is \$count\n";
use Interpolation 'count:\$->\$' => sub {
if (@_ == 2) {
\$count{\$_[0]} = \$_[1]
} else {
\$count{\$_[0]}++
}
};
# print "Current count of A is \$count{A}\n";

# Cool examples

SQL
use Interpolation "'" => sub {\$_ = \$_[0]; s/'/''/g; "'".\$_};
...
\$db->Sql("SELECT * FROM People WHERE LastName = \$'{\$lastname}'");

When passing strings to SQL you have to escape the apostrophes (and maybe some other characters) this crazy hack allows you do it quite easily.

Instead of "... = '\$variable'" you write "... = \$'{\$variable}'" et voila ;-) You may of course use this syntax for whatever string escaping you like.

# Warnings

It's easy to forget that the index to a \$hash{...} is an arbitrary expression, unless it looks like an identifier. There are two gotchas here.

Trap 1.
print "\$X{localtime}";

Here the X formatter is used to format the literal string localtime; the localtime built-in function is not invoked. If you really want the current time, use one of these:

print "\$X{+localtime}";
print "\$X{localtime()}";
Trap 2.
print "\$X{What ho?}";

This won't compile---you get `search pattern not terminated'. Why? Because Perl sees the ? and interprets it as the beginning of a pattern match operator, similar to /. (Ah, you forgot that ? could be a pattern match delimiter even without a leading m, didn't you?) You really need

print "\$X{'What ho?'}";

The rule is simple: That thing in the braces that looks like a hash key really is a hash key, and so you need to put it in quotes under the same circumstances that you need to put any other hash key in quotes. You probably wouldn't expect this to work either:

\$V = \$X{What ho?};

# Author

Now maintained by, Jenda Krynicky. http://Jenda.Krynicky.cz/#Interpolation